With the primary industry emphasis on autonomous fume hood controllers and the operation of the laboratory control loop in response to their causal laboratory airflow perturbations with the resultant non-optimum laboratory loop control lag, there exists a need for a more holistic approach. Current fume hood based controllers invoke perceived control actions without any concept or regard for laboratory or neighboring fume hood(s) flow history or perceived future demands. This approach and its resultant lag has in essence rendered laboratory controllers virtually impossible to setup and tune for any reasonably wide turndown range for any but the simplest (and mostly inaccurate) opening size based analog positioning systems.
If the laboratory is assumed to be the primary control loop and the individual hoods/sashes and exhaust dampers are assumed to be just a varied group of controllable exhaust vents, with make-up air loops on individual hoods (if applicable) and supply fans/dampers just another varied group of input vents, and laboratory differential pressure (inside lab/outside lab) an indication of room door(s) flow direction then perhaps a more straightforward and effective control strategy will result.
With a neuron-based approach, the generated network variables to/from all neurons will appear on the network with the resultant binding of certain variables for localized individual input/exhaust vent control and the binding of (laboratory) global variables for whole system control will result in a flatter, more timely architecture than what currently exists. With the binding of all local hood desired flow variables into a lab total exhaust demand variable at the same time as the local hood exhaust intelligent actuators are receiving the same desired flow variable will allow the lab control neuron (possibly the same product with a different superset application as the hood differential airflow sensor neuron since it too will measure differential pressure within/without the lab) to determine and effect the remaining damper/vents in conjunction with uncontrolled openings (doors et. al.) at essentially the same time.
In essence the command to the lab systems supply and auxiliary exhaust vents intelligent actuators will occur simultaneously with the command to the individual intelligent actuators bound to individual (hood) exhaust dampers. All the laboratory actuators will slew in concert with each other. The laboratory will truly be just one system with some local damper actuators (hoods) having certain minimums and maximums that must be maintained (actually no different than just about any flow controlling damper anyway). In addition, the automatic sash closure strategy is best handled at the laboratory level in a non-autonomous fashion.
Since the laboratory is the main application with localized subset applications, the inclusion of a meaningful/integral smoke/fire control strategy is as straightforward as the inclusion of the laboratory into the building HVAC/lighting control strategy. In essence, the present invention is a fully distributed architecture.
The concept of a fully distributed architecture of intelligent non-hierarchical input (sensors) and output (actuation) devices is predicated on the existence of a `flat architecture` in which intelligent devices act as peers communicating the sensible status of a process and implementing resultant controlling adjustments in response without the intervention of a superior hierarchical controller. In this architecture, inputs such as sash position sensors, air flow sensors, human presence sensors each has its own intelligence and communications capabilities in order to exist on a network with similarly intelligent and communication capable output devices such as intelligent dampers, intelligent VEnturi valves, intelligent actuators or intelligent sash closure devices such that complete process control actions can be implemented without the intervention or independent coordination of a `controlling element` which occupies a superior position in networking or calculation. In this `flat architecture` intelligent inputs and intelligent outputs act as peers in a non-hierarchial network and exchange sensed and controlling information in a configurable network which envisions single process loop control (such as a fume hood) and multiple process loop control (such as laboratories with multiple hoods) or arbitrarily configurable process loops to create `virtual applications`.
In this approach, the elements of input and output have the means of calculation or communication resident on them and integral to them. This method using distributed elements necessary to a process application where those elements can share a non-hierarchial high speed network in which a single control loop including one or more input or output, or more than one control loop which incorporates inputs and/or outputs which may be simultaneously used by more than one loop or calculation or application, allows the global sharing of sensed elements and the fuller, faster implementation of final control actions. Other approaches envision inputs and outputs that are digital or analog which are connected to a hierarchically superior `controller` which manages the inputs and outputs using their values as elements in calculations or electronics which determine final control actions. Subsequently, the `controller` may communicate with other `controllers` in a network of `controllers` which may share sensed or calculated values in order to accomplish global control strategies. In that approach, the inputs and output have no inherent or integral intelligence or communications capabilities.
This new approach increases the speed of communications, improves access to shared inputs and values, and removes significant cost elements required in hierarchical networks or global control applications.